Hands-On ROS for Robotics Programming

Selecting features in ML

Features in ML constitute a set of characteristics that have to be selected by the user, and it is this selection upon which the dataset is built. The expertise for making a good feature selection is more a question of experience and insight than a structured process. Hence, a good data scientist is one who understands the problem and can decompose it in its essential parts to find what the relevant features are. These act as the independent variables from which accurate predictions can be made.

To solve an ML problem, it is crucial to perform the right feature selection. If you do not detect the relevant features, no matter how much data you put in the solver, you will never get a good prediction. As shown in the ...